南京大学学报(自然科学版) ›› 2010, Vol. 46 ›› Issue (5): 520527.
于洪 ** , 李转运
Yu H ong, Li Zhuan Yun
摘要: 协同过滤是成功的个性化推荐技术之一. 但传统协同过滤算法由于不能及时反映用户的兴趣变化, 影响了推荐质量. 针对这个问题, 本文借鉴心理学上艾宾浩斯遗忘曲线来跟踪和学习用户的兴趣,
展开了协同过滤推荐算法的研究. 通过数学分析工具发现了与遗忘曲线拟合度较高的幂函数曲线, 并把用户的兴趣分为短期兴趣和长期兴趣, 提出了基于时间窗口的权重函数, 以此解决跟踪和学习用户兴
趣的难题. 结合项目的评分相似性和属性相似性来定义项目相似度数据权重函数. 将基于时间窗的数据权重与基于项目相似度的数据权重相结合来反应用户对项目的兴趣度. 最后, 在项目近邻模型的基础上
设计了跟踪用户兴趣变化的基于遗忘曲线的协同过滤推荐算法. 通过大量的实验工作确定了相关公式中系数的取值; 对比实验结果表明新的协同过滤推荐算法在推荐的准确性方面有显著的提高.
[ 1 ] Xu H L, W X, Li X D, et al. Comparison study of internet recommendation system. Journal of Software, 2009, 20(2) : 350~ 362. ( 许海玲, 吴潇, 李晓东等. 互联网推荐系统比较研究. 软件学报, 2009, 20( 2) : 350~ 362) . [ 2 ] Zeng C, Xing C X, Zhou L Z. Personalized service technology. Journal of Software, 2002, 13(2): 1952~ 1961. ( 曾?春, 邢春晓, 周立柱. 个性化服务技术综述. 软件学报, 2002, 13( 10): 1952~ 1961). [ 3 ] Zhao P, Geng H T, Wang Q Y, et al. Re search of a novel automatic personalized document recommendation system based on cluste ring and classification. Journal of Nanjing University(Natural Sciences), 2006, 42( 5) : 512~ 518. ( 赵鹏, 耿焕同, 王清毅等. 基于聚类和分类的个性化文章自动推荐系统的研究. 南京大学学报(自然科学) , 2006, 42( 5) : 512~ 518) . [ 4 ] Dong L, Xing C X, Wang K H. Based on different data sets of evaluating collaborative filte ring algorithms. Journal of Tsinghua University ( Science and Technology), 2009, 49( 4) : 590~ 594. ( 董丽, 邢春晓, 王克宏. 基于不同数据集的协作过滤算法评测. 清华大学学报( 自然科学) , 2009, 49(4): 590~ 594) . [ 5 ] Sarwar B, Karypis G, Konstan J, et al. Itembased collaborative filtering recommendation algorithms. Proceedings of the 10 th International Conference on World Wide Web, 2001, 285~ 295. [ 6 ] Tak-cs G, Pil-szy I, N?menth, et al. Scalable collaborative filtering approaches for large rec ommender system. Journal of Machine Learning Research, 2009(10) : 623~ 656. [ 7 ] Linden G, Smith B, York J. Amazon. com rec ommendations: Item?to?item collaborative filte ring. IEEE Internet Computing, 2003, 7( 1) : 76~ 80. [ 8 ] Das A, Datar M, Garg A. Google news personalization: Scalable online collaborative filtering. Proceeding of the WWW 2007/T rack: Industrial Practice and Experience. Banff, Alberta, Canada, 2007, 271~ 280. [ 9 ] Park S, Pennock D. Applying collaborative filtering techniques to movie search for better ranking and browsing. Proceedings of the 13 th Association for Cmputing Machinery Special Interest Group on Kniwledge Discovery in Data. San Jose, California, USA, 2007, 550~ 559. [ 10] Bell R, Koren Y. Improved neighborhood-based collaborative filtering. KDD ?Cup and Workshop at the 13 th Association for Cmputing Machinery Special Interest Group on Kniwledge Discovery in Data International Conference on Knowledge Discovery and Data Mining, 2007, 7~ 14. [ 11] Deshpande M, Karypis G. Item?based top-N recommendation algorithms. Association for Cmputing M achinery Transactions on Information Systems( TOIS) , 2004, 22( 1) : 143~ 177. [ 12] Xing C X, Gao F R, Zhan S N, et al. A Collaborative filtering recommendation algorithm incorporated with user interest change. Journal of Computer Research and Development, 2007, 44 (2) : 296~ 301. ( 邢春晓, 高凤荣, 战思南等. 适应用户兴趣变化的协同过滤推荐算法. 计算机研究与发展, 2007, 44(2): 296~ 301). [ 13] Zhang Y C, Liu Y Z. A Collaborative filtering algorithm based on time period partition. The proceeding of 3 rd International Symposium on Intelligent Information Technology and Security Informatics, 2010, 777~ 780. [ 14] Palam Software. ZGrapher?Graphing calculator software. http: / /www. palamsoft. com/ download. html, 2009- 11- 01. [ 15] Grouplens Research Lab. M ovieLens data sets. http: / /www. grouplens. org/node/73, 2009- 12- 20. |
No related articles found! |
|